thinQmaster Virtual testing partner Adaptive practice exams that find your weak spots and drill them. Every answer explained. From $69
Enroll

thinQmaster

Find your exam

Five questions free, before you decide.

Step 1 · choose a vendor

In one of our camps? Your practice exams come with the tuition — they are on your study page already.

AI & Machine Learning · Intermediate

Data Leakage

When information unavailable at prediction time leaks into training, producing unrealistically good evaluation results.

Definition

Leakage occurs through target-derived features, preprocessing fitted on the full dataset before splitting, duplicate records across splits, or random splits on time-ordered data. Leaky models look excellent in validation and fail in production.

Example

Including a refund_issued flag when predicting whether an order will be returned.

Exam tip

Suspiciously high validation accuracy is a hint to look for leakage.

Related terms